The Masakhane community is developing a comprehensive, open, community-driven Data Annotation Playbook for high-quality datasets across diverse languages, domains, and modalities. We invite researchers, practitioners, and communities to submit chapter proposals — globally welcomed, with a focus on African and low-resource language contexts.
We welcome chapter proposals covering one or more data modalities. Submissions may focus on a single area or span multiple.
Written-language tasks across classification, generation, retrieval, and structured prediction.
Voice and audio annotation for recognition, synthesis, and paralinguistic understanding.
Image, video, and cross-modal grounding with strong African-context relevance.
Submit a short proposal (max 5 pages per chapter) outlining the planned chapter contents, structure, and your relevant experience. Multi-modality submissions count separately.
Selected proposals are invited to develop the full chapter following editorial guidelines. Final chapters that meet the guidelines and align with the proposal are published.
Chapters should prioritize clarity, usability, and reproducibility, and follow best practices over purely theoretical discussion.
Clear, structured annotation guidelines that can be picked up and used without prior context.
From task definition through annotation to quality assurance — covering the full lifecycle.
Explicit considerations for ethics, bias, and inclusivity, with practical mitigations.
Practical tools, templates, examples, and (where possible) real-world case studies.
To support contributors, each accepted chapter receives an honorarium of USD $1,000, paid after the final chapter is submitted and meets the editorial guidelines.
Exact dates will be announced shortly. Subscribe to the AfricaNLP newsletter to be notified.